Both increases of 2 or more hours a day of sleep between the years 1986 and 2000 and consistent sleep duration of 6 hours a day or less were associated with an increased risk for subsequent type 2 diabetes, but the short sleep risk lessened after adjustment for confounders.